Please forgive my ignorance. Why do the colors, especially the shadows, look so different depending on the sharpening method? I thought the only thing that changed was the sharpening, with more details or aggressive sharpening in the case of XD2s, but this is not realistic, the colors are totally off. It’s also definitely possible that I’m not using the right type of sharpening for the image, I’m just going by the description of “ultimate quality and extra details”, it doesn’t say anything about colors. Taken with a Nikon D750, ISO 100, 135 seconds if that means anything to you. Thanks!
I think that’s DeNoise not sharpening; It’s trying to fill in the noise in the grass with more green, and over-saturating it a little.
I would use a mask on the grassy area, then apply DeNoise and adjust the luminance (slider right underneath) to see if that fixes the issue. You may also want to tweak the exposure in that mask.
I am not sure what is happening here, but I have used DP 3 and XD2s on hundreds of pictures and have never seen any significant change in colors between those two noise algorithms.
Mark
You have obviously lifted the shadows. Just check settings under another menu …
Referring to your screenshots you used different denoising, but otherwise the same settings.
.
… I’m just going by the description of “ultimate quality and extra details” …
You must not cling on all marketing blubb. 
Compared to the newer DP 3 / XD3 X-trans methods, which handle CA correction differently (!), DP XD/XD2s pulls out more details by default.
If the latter looks unnatural → simply adjust the settings.
Additionally you can increase the CA corrections to check if the color rendition improves (otherwise, use DP 3).
From PL 9 onwards you can adjust the general denoise settings locally
( see @LVS’ proposal ).
@Gabe Can you share the original file and the .dop sidecar?
I’ts an interesting case that I’ve not seen before. Some tinkering might help us find where the change really happens. Sometimes, It’s as simple as having an overlay of a low-light warning, but that is not the case here as far as I can see.
Also, the screens don’t tell the whole story because of the filter that is applied. If you unselect the “Details” switch and flip the “Active corrections” switch instead, all applied corrections - and possibly the source of the issue - will be shown.
1 Like
I can’t recreate this effect. Are you sure the only difference is your choice of denoising algorithm?
As already requested by @platypus, can you share the original file and .dop sidecar?
1 Like
I’ll upload the files as soon as I get home.
1 Like
It’s the exact same settings, the color changes happen just by toggling the sharpening method.
2023-11-15.zip (21.0 MB)
@platypus @stuck Here’s the .nef and .dop files. The color shift becomes apparent as you zoom in to 75-100% (at least for me). I’m also uploading a couple of .jpgs to show that the color change persists on export; I thought it was just a preview issue but it’s more than that (I know, jpgs are not the best, but they’re good enough to illustrate this. If I upload the .dng files, they’re much bigger. Feel free to do your own exports).
Thanks!
I wouldn’t have bought PL9 to begin with, if I had thought of that. Rookie mistake.
Also, I just disabled Lens Sharpness Optimization, CA and all the other settings I could think of and the only thing that remains is the denoising method itself that does something different in DP3 than it does in XD. Look at these two screenshots, the only difference is the Denoising switch.
It’s not grass, it’s actually pebbles and sand. So no green there to begin with.
View and compare on 100%. With 75% or below you might see strange colors.
George
Changing the noise reduction method also changes the demosaic formula, which can affect color. Sometimes (particularly with fine detail) DeepPRIME methods produce false color in the results. Fortunately it’s fairly uniform in your example, and you have numerous options for correcting the problem. You might also want to provide the sample image file to DxO support (support.dxo.com) - they can evaluate the image processing and see if it can be improved for greater color accuracy.
Thank you. I still can’t recreate the colour shift though, even when viewing at 100%.
I opened your image as it is (= with DP 3), added a Virtual Copy (= changed it to DP XD/XD2s) and compared them in split preview mode.
Your version (Master) is on the left and the VC on the right.
→ 20231115-18-39-18-DSC_7560.NEF.dop (20,5 KB)
The comparison shown was created using a monitor in sRGB color mode, which offers the best possible color reproduction for this forum (which doesn’t support color profiles).
I also checked with the monitor set to AdobeRGB and Native (the maximum my calibrated CG2730 can display), and the result was the same – no green tint in VC 1 (DP XD/XD2s).
I experimented with an additional VC where I disabled chromatic aberration correction as well maximized the settings – but there was no (or hardly any) difference compared to VC 1 (DP XD/XD2s). Then I deleted that VC.
So I’m wondering why you see this green tint in the pebbles.
.
Curious about the difference in noise reduction between DP 3 and the DP XD/XD2s mode, I zoomed in significantly (even using the magnifying glass) to 200% and 400% and noticed a slightly different level of detail, but nothing really worth mentioning.
You took the photo at ISO 100, and this old camera (I still have the same one) is excellent when it comes to noise reduction.
Normal editing gives expected results here.
The only way I could reproduce your screenshot was to change masks overlay color to something green and use AI mask with selection displayed.
1 Like
I could only make the peebles go green by masking them and then tweaking the HSL wheel.
I just opened a ticket with DxO, fingers crossed they’ll find something. And I’ve tried other images and it does the same.
